News media bias significantly influences public perception and trust in journalism. As news media outlets play a critical role in shaping public opinion, detecting bias in reporting is essential to ensure balanced and fair communication. This study presents an unsupervised machine learning framework for identifying media bias, leveraging a combination of natural language processing (NLP) techniques with statistical analysis. Large-scale news corpora are evaluated across multiple parameters, including content, tone, emotion, readability, and balance to uncover patterns of bias. The choice of an unsupervised machine learning approach serves the objective to address the challenge of the unavailability of a gold standard labelled dataset. The proposed system demonstrated its effectiveness by analyzing the four most prominent Indian media outlets: The Times of India, Hindustan Times, Deccan Herald, and India Today. Experimental results from these sources showcase the system’s effectiveness in detecting and differentiating bias levels across the selected parameters, offering valuable insights into the landscape of media reporting in India.